Hassaku aims to be a model with a bright, clear anime style. Model focus are nsfw images, but also with a high emphasis for good looking sfw images as well. My <a target="_blank" rel="ugc" href="https://discord.gg/zSR5FcYWWE">Discord</a> for everything related to anime models and art. My <a target="_blank" rel="ugc" href="https://discord.gg/zSR5FcYWWE">Discord</a> for everything related to anime models and art. If you'd like to support my work, you can do so at: <a target="_new" rel="ugc" href="https://subscribestar.adult/citrus-models">SubscribeStar</a>, every bit helps and is truly appreciated!____________________________________________________________ Supporters:Thanks to my supporters tamashiicolle, pttcot, SETI and Kodokuna_____________________________________________________Using the model:Use mostly <a target="_blank" rel="ugc" href="https://danbooru.donmai.us/">danbooru</a> tags. No extra vae needed. For better promting on it, use this <a target="_blank" rel="ugc" href="https://aituts.com/novelai-anime-prompt-techniques/">LINK</a> or <a target="_blank" rel="ugc" href="https://lunarmimi.net/freebies/novelai-anime-girl-prompt-guide/#1basic">LINK</a>. But instead of {}, use (), <a target="_blank" rel="ugc" href="https://github.com/AUTOMATIC1111/stable-diffusion-webui">stable-diffusion-webui</a> use (). Use "masterpiece" and "best quality" in positive, "worst quality" and "low quality" in negative.My negative ones are: (low quality, worst quality:1.4) with extra monochrome, signature, text or logo when needed.Use a clip skip 1 or 2. Clip 2 is better for private parts, img2img and prompt following. Clip 1 is visually better, because i assume, the model has more time and freedom there. I use clip2.Don't use face restore and underscores _, type red eyes and not red_eyes.Don't go to really high resolutions. Every model, like hassaku, get lost in the vastness of big images and has a much higher chance to greate, as example, a second anus._____________________________________________________<a target="_blank" rel="ugc" href="https://rentry.org/hdgpromptassist#loras">Loras</a>:Every LoRA that is build to function on anyV3 or orangeMixes, works on hassaku too. Some can be found <a target="_blank" rel="ugc" href="https://rentry.org/hdglorarepo">here</a>, <a target="_blank" rel="ugc" href="https://gitgud.io/gayshit/makesomefuckingporn#lora-list">here</a> or on civit by <a target="_blank" rel="ugc" href="https://civarchive.com/user/lottalewds">lottalewds</a><a target="_blank" rel="ugc" href="https://civarchive.com/user/fuggy">, </a><a target="_blank" rel="ugc" href="https://civarchive.com/user/Trauter">Trauter, </a><a target="_blank" rel="ugc" href="https://civarchive.com/user/Your_computer">Your_computer</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/user/ekune">ekune</a> or<a target="_blank" rel="ugc" href="https://civarchive.com/user/fuggy"> </a><a target="_blank" rel="ugc" href="https://civarchive.com/user/lykon">lykon</a><a target="_blank" rel="ugc" href="https://civarchive.com/user/fuggy">.</a>_____________________________________________________Black result fix (vae bug in web ui): Use --no-half-vae in your command line argumentsI use a Eta noise seed delta of 31337 or 0, with a clip skip of 2 for the example images. Model quality mostly proved with sampler DDIM and DPM++ SDE Karras. I love DDIM the most (because it is the fastest).

Hassaku (SD1.5)

Small update on those things: <ul><li>Slightly bigger breasts in compare on prompts large and huge breasts on some images</li><li>Slightly better formed hands midrange (from bad hands to slightly better bad hands... on some images it helps)</li><li>Better response on prompt fellatio with a higher probability, that the tip is in the mouth</li><li>Details on clothes or bodys (like wings) are less likely to be detached from bodys and is more equal right and left</li><li>Less extra limbs, ears, animal ears, wings</li><li>Seems that details on clothes are less random formed out</li><li>Slightly better constructed backgrounds</li></ul> Hard to determine: <ul><li>Waists seems more norrow, not sure about that</li></ul> What is worse: <ul><li>On some prompts, clothes apear more lose with more "strings" or holes all over the place Be aware that those points are tested over many images on clip2, it can be, that on some images, those points don't do anything or are even worse than before. It is hard to update things with training, if most training material don't look like the images my model itself produce. Need to merge back and forth, so the model itself looks and behaves like before training. It have less changes with training, because of it. Model trained on 150 fellatio images and later merged back with hassaku and a bit of koji. Archived a bit more stability with add difference (on lower resolution).</li></ul>

hassakuSD15_v13.safetensors

hassakuHentaiModel_v13 (latest as of 30 June 2023)

Hassaku (hentai model) V1.3

v1.3

V1.2

V1.1

V1to1.3-inpainting

Hassaku

hassakuHentaiModel_v13.safetensors

Hassaku aims to be a model with a bright, clear anime style. Model focus are nsfw images, but also with a high emphasis for good looking sfw images as well. My <a target="_blank" rel="ugc" href="https://discord.gg/zSR5FcYWWE">Discord</a> for everything related to anime models and art. My <a target="_blank" rel="ugc" href="https://discord.gg/zSR5FcYWWE">Discord</a> for everything related to anime models and art. If you'd like to support my work, you can do so at: <a target="_new" rel="ugc" href="https://subscribestar.adult/citrus-models">SubscribeStar</a>, every bit helps and is truly appreciated!____________________________________________________________ Supporters:Thanks to my supporters tamashiicolle, pttcot, SETI and Kodokuna_____________________________________________________Using the model:Use mostly <a target="_blank" rel="ugc" href="https://danbooru.donmai.us/">danbooru</a> tags. No extra vae needed. For better promting on it, use this <a target="_blank" rel="ugc" href="https://aituts.com/novelai-anime-prompt-techniques/">LINK</a> or <a target="_blank" rel="ugc" href="https://lunarmimi.net/freebies/novelai-anime-girl-prompt-guide/#1basic">LINK</a>. But instead of {}, use (), <a target="_blank" rel="ugc" href="https://github.com/AUTOMATIC1111/stable-diffusion-webui">stable-diffusion-webui</a> use (). Use "masterpiece" and "best quality" in positive, "worst quality" and "low quality" in negative.My negative ones are: (low quality, worst quality:1.4) with extra monochrome, signature, text or logo when needed.Use a clip skip 1 or 2. Clip 2 is better for private parts, img2img and prompt following. Clip 1 is visually better, because i assume, the model has more time and freedom there. I use clip2.Don't use face restore and underscores _, type red eyes and not red_eyes.Don't go to really high resolutions. Every model, like hassaku, get lost in the vastness of big images and has a much higher chance to greate, as example, a second anus._____________________________________________________<a target="_blank" rel="ugc" href="https://rentry.org/hdgpromptassist#loras">Loras</a>:Every LoRA that is build to function on anyV3 or orangeMixes, works on hassaku too. Some can be found <a target="_blank" rel="ugc" href="https://rentry.org/hdglorarepo">here</a>, <a target="_blank" rel="ugc" href="https://gitgud.io/gayshit/makesomefuckingporn#lora-list">here</a> or on civit by <a target="_blank" rel="ugc" href="https://civitai.com/user/lottalewds">lottalewds</a><a target="_blank" rel="ugc" href="https://civitai.com/user/fuggy">, </a><a target="_blank" rel="ugc" href="https://civitai.com/user/Trauter">Trauter, </a><a target="_blank" rel="ugc" href="https://civitai.com/user/Your_computer">Your_computer</a>, <a target="_blank" rel="ugc" href="https://civitai.com/user/ekune">ekune</a> or<a target="_blank" rel="ugc" href="https://civitai.com/user/fuggy"> </a><a target="_blank" rel="ugc" href="https://civitai.com/user/lykon">lykon</a><a target="_blank" rel="ugc" href="https://civitai.com/user/fuggy">.</a>_____________________________________________________Black result fix (vae bug in web ui): Use --no-half-vae in your command line argumentsI use a Eta noise seed delta of 31337 or 0, with a clip skip of 2 for the example images. Model quality mostly proved with sampler DDIM and DPM++ SDE Karras. I love DDIM the most (because it is the fastest).

Files with SHA256: 7eb674963a0d93fcead95905b4d658d1e94e9a36230c0c732f65c3f53d8888aa